Introduction

With the modernization of the world, millimeter-wave (mmWave) frequencies are being rapidly adopted to enable high bandwidth and low latency communication1. Military satellite applications can be presented as an example of this transition to the mmWave frequencies. For instance, military satellites have typically focused on the C-band (uplink: 5.8- 6.7GHz and downlink: 3.4- 4.2GHz) and X-band (Uplink: 7.9- 8.4GHz and downlink: 7.25- 7.75GHz) operations. However, over the years, the increasing demand for bandwidth from unmanned aerial systems (UAS) in the same bands facilitated the transition to higher frequencies, such as the Ku and Ka bands2. Similarly, the Next-G wireless communication systems will also depend on the mmWave frequencies to establish ultra-low latency data channels3,4,5,6. In 6G communication, the FR3 band and Q, E, W, D, etc., are considered to provide ultra-wide bandwidth and extremely high data transfer rate7,8,9,10. The applications of these frequencies can be seen in many other areas as well. Radar, satellite, radio astronomy, security screening, imaging systems, etc., are some additional examples11,12,13,14,15,16,17,18.

The transition to the mmWave bands still faces transmission range issues due to the high path loss at such high frequencies19,20. Therefore, the use of low frequencies is obvious for long-range communication21,22. That is why sub-6 GHz band frequencies are kept as an integrated part of all modern communication systems23. In satellite communication, these low frequencies are very important, as the distance between the transmitter and receiver is hundreds of kilometers. The modern Direct-Satellite-to-Cell technology is one instance where the communication uses sub-6 GHz band frequencies24,25. As the use of diverse frequency bands rapidly increases, integration of transceiver technology of various sizes, frequencies of operation, and bandwidths is necessary.

A single antenna with an extremely wide bandwidth could be an option to enable both the low and high frequencies. Ultra-wideband (UWB) Yagi-Uda, Vivaldi, tightly coupled arrays (TCA), etc., are some examples of such antennas26,27,28,29. However, the realized gain performance of UWB antennas across the large operating frequencies is unsatisfactory. In addition, most of these antennas, in particular the TCAs and Vilvaldi, suffer from the height requirement, which restricts low-profile applications. Aside from antenna constraints, UWB RF front-end hardware suffers from poor linearity and high losses and is cost-prohibitive. Hence, a co-integrated antenna optimized at the frequencies of interest is a better possible solution to alleviate the bandwidth requirement of RF front-end hardware without sacrificing the low-profile feature.

Indeed, a co-integrated or shared aperture approach is beneficial when widely separated frequencies are required to integrate into a single system30. The elements for each frequency are inherently isolated from the others, which decreases crosstalk among the antenna elements31. Both single-feed and multi-feed shared aperture antennas using single-element or array structures can be found in the literature32,33,34,35. When the antennas are fed individually, a multiplexer of a filter bank is not required; hence, the front-end design becomes simple. In this work, we present a multiband co-integrated design using seven different antennas, as depicted in Fig. 1, covering widely used frequencies from UHF to Ka-band. This antenna enables operation across 2G, 3G, 4G, and 5G cellular bands, satellite communication frequencies, and military bands. Notably, the sub-6 GHz antennas are fed separately to ease front-end design. The mmWave frequencies offer beam-steering capabilities to support advanced 5G/6G technologies. Therefore, the presented antenna system—with a total size of 200 mm \(\times\) 85 mm—is well suited for commercial applications such as fixed wireless access (FWA) customer premises equipment (CPE), vehicle-mounted communication systems, small cells, and integrated access and backhaul (IAB) nodes, as well as military platforms requiring cellular, satellite, and multiband connectivity. The organization of the paper is as follows. Section II of the article describes the design architecture of the co-integrated antenna. The mmWave antenna designs are outlined in Section III. The fabricated prototypes and measurement results are reported in Section IV, and the concluding remarks are in Section V.

Fig. 1
figure 1

Co-integrated sub-6 GHz and mmWave antennas on a single aperture.

Full size image

Design of the co-integrated antenna

Figure 1 shows the co-integrated antenna operating at seven different bands ranging from UHF to Ka-band. These bands are divided into six groups based on the commercially used cellular frequency bands (viz. 2G, 3G, 4G, 5G, and some of the 6G), as listed in Table. 1. For each of these bands, an antenna or an antenna array is optimized separately and integrated on an 8.5 cm \(\times\) 20 cm shared aperture. The design details of each antenna element or array are shown in Fig. 2 and Fig. 3. As many antennas are placed on a single aperture, proper placement planning is required to maintain a low level of crosstalk among the antennas.

Table 1 Frequency Groups for the Design Plan.
Full size table

Design of the sub-6 GHz antennas

Considering the available area on the aperture, five antennas in the sub-6 GHz range are designed separately. After optimizing independently, the antennas are placed on the aperture, as depicted in Fig.1. We employed four inverted-L (inv-L) antennas and one slotted patch antenna to achieve compact size, wideband operation, and controlled interference. Studies show that at low frequencies, inv-L antennas are narrowband; hence, parasitic elements are used to enhance the bandwidth.

The inv-L antenna resonates at approximately \(\lambda /4\) of the target frequency, making it compact but naturally narrowband. To enhance its impedance bandwidth, parasitic arms slightly longer than \(\lambda /4\) were introduced, leveraging mutual coupling to broaden the operational band. For the lowest band (G1: 600–900 MHz), additional tuning can be achieved using matching circuits with inductors and capacitors, a method commonly used in cellular handsets. The choice of radiator geometry directly impacts the S-parameters and gain, as inv-L elements tend to show shifts in matching and more omnidirectional gain patterns, while the slotted patch yields broader bandwidth and more stable directional gain. The placement of the sub-6 GHz antennas was optimized through positional rearrangement during the layout process to minimize mutual coupling and electromagnetic interference. While increased spacing between elements could have further reduced interaction, the limited board area imposed significant constraints, and the current configuration reflects the most effective trade-off between isolation and spatial efficiency. Additionally, antenna size at lower frequencies posed another design challenge. To accommodate the shared aperture within the restricted footprint, a meandering technique was employed to reduce the lateral dimension of the inverted-L antenna operating at the G1 band, as illustrated in Fig. 2. The length of the main arm of the inv-L antenna is 153 mm, and the coupled arm is 90 mm. Using meandering, a 40 mm reduction in the lateral direction is obtained along the length of the aperture. The width of the main arms is 4 mm, and the parasitic arm and meandered portion of the main arm are 1 mm.

Fig. 2
figure 2

Various antenna elements designed and optimized for operation at different frequency bands.

Full size image
Fig. 3
figure 3

MmWave antenna array design at 39 GHz.

Full size image

The main arm lengths for the following two groups, G2 and G3, are 70 mm and 53 mm, with a 4.5 mm width, respectively. The parasitic arm lengths are 41 mm and 33 mm, respectively, and the width is 2 mm for both. For G4, a slotted patch antenna is used to ensure optimal placement that reduces crosstalk among the antenna elements. The patch size is 26 mm \(\times\) 24 mm, and the slot length is 45 mm. For the G5 group, we use another inv-L antenna with a 35 mm arm length. Rogers 4003 C with a dielectric constant of 3.55 and a height of 1.52 mm is used as the substrate for all cases.

Millimeter-wave antennas designs

For the mmWave group, two 4\(\times\)4 patch arrays are designed to operate at 28 GHz and 39 GHz, respectively. The arrays are integrated with the sub-6 GHz antennas on a shared platform to ensure compactness. For prototyping, the mmWave arrays were fabricated and measured separately on different substrates with independent ground planes. They were manually assembled onto the full structure using temporary taping for initial validation. The patch-array geometry was chosen for its ability to provide high gain, controlled radiation patterns, and potential for beam steering. The elements are spaced at half of the free space wavelength, which ensures constructive aperture illumination and reduces grating lobes to support beamforming functionality in future implementations. This geometry directly influences the gain (through aperture efficiency) and S-parameters (through inter-element coupling and feed matching).

The mmWave antenna arrays were designed with element spacing approximately equal to half the guided wavelength to support beamforming functionality in future implementations. In this work, both designs are optimized for boresight radiation, as depicted in Fig. 2 (bottom, right) and Fig. 3. An aperture-coupled feeding network is designed and implemented for each of the designs. Aperture coupling is achieved using four simple transmission lines. By applying proper phase shifts at the input ports of these transmission lines, one can enable 1D beam scanning. For testing purposes, the ports are replaced with a corporate feed. In this case, only boresight radiation can be achieved. The optimized length and width of the 28 GHz antenna are 3 mm and 1.98 mm, respectively. Rogers RO4003C (\(\varepsilon _r\)=3.55) with 0.203 mm height is used as the bottom substrate, and RT/duroid 5880 (\(\varepsilon _r\)=2.2) with height 0.5 mm is used as the top substrate. For the 39 GHz array design, the optimized patch size is 3.04 mm \(\times\) 2.03 mm. The top substrate is chosen to be Rogers RT/duroid 5880 (\(\varepsilon _r\)=2.2) with a height of 0.5 mm, and the bottom substrate is RO3006 (\(\varepsilon _r\)=6.15) with a height of 0.25 mm. Compared to the inv-L and slotted patch antennas used at sub-6 GHz, the mmWave patch arrays exhibit significantly higher directional gain and lower impedance sensitivity, making them ideal for high-frequency operation where path loss is severe. It should be noted that, for prototyping purposes, the mmWave arrays were fabricated on separate substrates and then manually placed onto the full structure. This process required minor modifications to the ground plane, which unintentionally introduced a defected ground-like feature. While such a feature may have contributed positively to reducing mutual coupling, it was not part of the original design intent. In future versions, the arrays will be fully integrated into a multilayer structure with a continuous ground plane to eliminate any ambiguity.

Simulation of the seven-band co-integrated aperture

After designing and optimizing the individual antennas, the elements were placed on a single aperture. Fig. 1 shows the combined antenna containing all the elements. The simulation of the structure, including all seven antennas, is very important as the crosstalk can seriously hamper the antenna performance. The mmWave antenna arrays are operating at widely separated frequencies. As a result, the coupling between those two arrays is minimal, and almost no coupling is expected from and to the sub-6 GHz antennas.

To evaluate the crosstalk among the antennas, the surface currents are plotted by activating the antennas sequentially. We first plot the simulated S-parameters of each isolated antenna and compare them to those when the same antenna is integrated on a single aperture. Fig. 4 (a) shows the simulated result of the \(G_1\) inv-L antenna (600 MHz - 900 MHz). The \(S_{11}\) plots show a slight shift, which is mainly due to the coupling to the nearest antennas and the larger platform size of the aperture. We note that, in the plot, \(S_{21}-G_n\) refers to the coupling between the active antenna and the \(G_n\) antenna, where \(n=2,3,4,5\). The plot shows that the crosstalk level to the other sub-6 GHz antennas is below −25 dB.This indicates that integration into the larger aperture actually stabilizes the impedance matching of G1, explaining why it performs slightly better when combined compared to individually. The surface current plot shown in Fig. 4 (b) further confirms the minimal coupling to the neighboring antennas. The simulated responses obtained from the next group (\(G_3\)) are depicted in Fig. 5 (a). The antenna shows a little change in the resonance frequency compared to the single antenna results, but retains the wideband matching characteristics. The coupling to the other antennas is very low, except for the closest neighbor, which resonates at 2200 MHz. The \(S_{21}\) magnitude is slightly above −15 dB for frequency bands greater than 1700 MHz. The surface current plot explains the reason for such a crosstalk level. Looking at Fig. 5 (b), one can see that the parasitic arm of the \(G_3\) antenna is coupling to the \(G_2\) antenna, implying reduced isolation. The isolation could be improved by swapping the positions of the \(G_2\) and \(G_3\) antennas. However, this could not be done due to the limited available space on the structure. The S-parameters and the surface current plot for \(G_3\) are shown in Fig. 6 (a) and (b), respectively. We note that the coupling resulted in \(G_3\) having a slightly narrower bandwidth. In addition to the placement constraint, the same order of length of the parasitic arms of this antenna and the G5 antenna caused changes in the responses. A little coupling is also present in the G1, but the effect is negligible due to the large frequency difference. The placement of the \(G_4\) antenna operating (3500 MHz - 3700 MHz) on the bottom side of the aperture resulted in improved isolation, as depicted in Fig. 7 (a) and (b) with a minor change in the \(S_{11}\) response. The S-parameters and the surface current plot for the \(G_5\) antenna are shown in Fig. 8 (a) and (b), respectively. In both the separate and combined cases, the antenna showed wideband matching followed by a resonance shift toward the lower frequency. A slight degradation of the crosstalk performance is observed as the antenna is in the middle of the board. However, the coupling level is near −20 dB, which is satisfactory for such a compact design.

Figure 9 [(a)-(e)] and Fig.10 [(a)-(b)] show the 2D radiation patterns of the inverted-L antennas and the patch antennas operating separately in the sub-6 GHz bands and mmWave bands respectively. (Fig. 11)

Fig. 4
figure 4

(a) S-parameter simulation of \(G_1\) antenna separate vs. integrated on the aperture and (b) surface current plot at 750 MHz. \(S_{21}-G_n\) refers to the coupling between the active antenna and the \(G_n\) antenna.

Full size image
Fig. 5
figure 5

(a) S-parameter simulation of \(G_2\) antenna separate vs. integrated on the aperture and (b) surface current plot at 1700 MHz.

Full size image
Fig. 6
figure 6

(a) S-parameter simulation of \(G_3\) antenna separate vs. integrated on the aperture and (b) surface current plot at 2200 MHz.

Full size image
Fig. 7
figure 7

(a) S-parameter simulation of \(G_4\) antenna separate vs. integrated on the aperture and (b) surface current plot at 3500 MHz.

Full size image
Fig. 8
figure 8

(a) S-parameter simulation of \(G_5\) antenna separate vs. integrated on the aperture and (b) surface current plot at 5200 MHz.

Full size image
Fig. 9
figure 9

Simulated 2D radiation pattern(Co and Cross pol) of single element at (a) G1: 760MHz, (b) G2: 1.7GHz, (c) G3: 2.25GHz, (d) G4: 3.6GHz, (e) G5: 4.5GHz.

Full size image
Fig. 10
figure 10

Simulated 2D radiation pattern(Co and Cross pol) of mmWave array at (a) G6: 28GHz and (b) 38.6GHz.

Full size image
Fig. 11
figure 11

Simulated efficiency on integrated aperture across Sub-6 GHz bands and mmWave bands.

Full size image

Fabrication and measurement of the seven-band co-integrated antenna

A prototype for the seven-band co-integrated antenna is fabricated in-house using an LPKF Protolaser U4. The mmWave antenna prototypes are fabricated separately and then integrated into the shared structure, as depicted in Fig. 12. Specifically, the multilayer structures of the mmWave antennas are carefully stacked and glued together to avoid air gaps and ensure the structural stability of the antenna arrays. For each of the arrays, a 2.4 mm coaxial connector is used for feeding. Alternatively, conventional SMA connectors are used to feed the sub-6 GHz antennas. Fig. 12 (a) shows the in-house produced prototype’s top and bottom views and (b) shows the VNA measurement setup for G4 antenna. The bottom view shows the partial ground plane, which is intentionally considered during the design process to obtain omnidirectional coverage at the sub-6 GHz frequencies.

Fig. 12
figure 12

Fabricated prototype of the seven-band co-integrated antenna covering 600 MHz to 900 MHz, 1500 MHz to 1800 MHz, 1800 MHz to 2700 MHz, 3500 MHz to 3700 MHz, 4700 MHz to 6000 MHz, 28 GHz, and 39 GHz, (b) VNA measurement set up for 3500MHz to 3700MHz band.

Full size image

S-parameters measurement

A Vector Network Analyzer (VNA) is used to measure the S-parameters of the sub-6 GHz and mmWave antennas. Figure 13 shows the measured S-parameters of all sub-6 GHz antennas. We observe that the \(S_{11}\) plots between simulation and measurements are in great agreement for the majority of the sub-6GHz antennas. The simulation setup of the large aperture was a bit troublesome due to the various sizes of the radiating structures, which may have contributed to the discrepancies. Moreover, fabrication and measurement tolerance may act as additional factors. Nevertheless, the overall performance is very satisfactory. In particular, the measured responses for inv-L \(G_1\) antenna are shown in Fig. 13 (a). A wideband matching is observed in the measured \(S_{11}\) plot. The slight discrepancy with the simulation is mainly due to fabrication errors. The measured crosstalk level is very low and is comparable to the simulation (see Fig. 4 (a)). Figure 13 (b) shows the simulated and measured results for the \(G_2\) antenna. The measured coupling magnitude is below −20 dB, except for the coupling with the \(G_3\) antenna, which is expected. These results closely match the simulations in Fig. 5 (a). The measured responses obtained for \(G_3\) are plotted in Fig. 13 (c). The results show a very good match in the desired frequency range. The crosstalk performance is comparatively degraded, though it is still satisfactory. This was also observed in the simulation results in Fig. 6 (a). Figure 13 (d) shows the measured responses of the \(G_4\) slotted patch. The measured matching shows a wider bandwidth compared to the simulated matching. The measured \(S_{21}\) values are in excellent agreement with the simulated responses (see Fig. 7 (a)), confirming a very low level of coupling magnitude. The last antenna responses from the sub-6 GHz are shown in Fig. 13 (e). Wideband matching is observed from the measured \(S_{11}\) response. The measured coupling magnitudes are also very low (below −20 dB).

Fig. 13
figure 13

Measured S-parameters when (a) \(G_1\) antenna is active, (b) \(G_2\) antenna is active, (c) \(G_3\) antenna is operating, (d) \(G_4\) antenna is operating, and (e) \(G_5\) antenna is operating. Simulated \(S_{11}\) values are added for comparison purposes. \(S_{21}-G_n\) refers to the coupling between the active antenna and the \(G_n\) antenna.

Full size image

The measured responses for the 28 GHz and 39 GHz antennas are compared with the simulation results in Fig. 14 (a) and (b), respectively. Although an excellent \(S_{11}\) performance is observed for the 39 GHz antenna, a slight shift in the 28 GHz antenna resonance is observed. The reason behind the shift might be the complex fabrication and prototype assembly. Notably, a minor misalignment between the upper and lower substrate can cause a massive difference in the response at such high frequencies.

Fig. 14
figure 14

Comparison of the simulated and measured S-parameters at 28 GHz and 39 GHz.

Full size image

Field measurement

Next, the radiation performance is measured. The StarLab anechoic chamber (range: 650 MHz to 18 GHz) is used to measure the sub-6 GHz bands, and the MVG \(\mu\)Lab anechoic chamber (range: 18 GHz to 110 GHz) is used to measure the 28 GHz and 39 GHz antennas. Figure 15 shows the antenna measurement setup in two different anechoic chambers. Figure 16 compares the simulated and measured radiation responses of \(G_1\). An excellent performance is observed for the frequency vs. gain plot (Fig. 16 (a)). Figure 16 (b) and (c) show the simulated and measured 3D radiation patterns, respectively. The measured 3D pattern has a slightly distorted shape, which may be caused by some measurement inaccuracy. The simulated and measured results for the inv-L \(G_2\) antenna are depicted in Fig. 17. Here, an excellent radiation performance is obtained. The frequency vs gain plot shows little discrepancy after 1600 MHz, but the overall gain magnitude is satisfactory. The gain pattern for the \(G_3\) group, i.e., 1800 MHz to 2700 MHz, is displayed in Fig. 18 (a), (b), and (c). Here, a perfect match is observed between the measured and simulated frequency vs. gain plot. A little discrepancy is observed in the 3D plots due to the positioning of the antenna, as described above. The frequency vs. gain plot for the G4 antenna is depicted in Fig. 19 (a), and the 3D gain patterns are shown in Fig. 19 (b) and (c). An excellent agreement is obtained for both comparisons. The measured responses from the last group of the sub-6 GHz antenna are recorded in Fig. 20 (a), (b), and (c). Little difference in the frequency vs. gain plot values and in the 3D patterns is observed. We expect that the fabrication and measurement tolerance may have caused such discrepancies in the results. Lastly, in Fig. 21, the measured radiation performance was compared with the simulated responses for the 28 GHz and 39 GHz antennas. A perfect match is observed between the measured and simulated radiation patterns. However, the measured gain value shows some loss compared to the simulated results. The in-house fabrication and prototype steps may introduce additional losses. Multiple layers of the antenna are very sensitive to positioning, as the exact coupled port is necessary to avoid performance alteration. Manual positioning and gluing of the layer may have introduced additional losses that were not accounted for during the simulations.Table 2, shows the comparison of our proposed antenna with other multibad antennas operating in multiple frequency spectrum30,36,37,38,39.

Fig. 15
figure 15

Field measurement of the sub-6 GHz and mmWave antennas. The StarLab anechoic chamber (range: 650 MHz to 18 GHz) is used to measure the sub-6 GHz bands, and the MVG \(\mu\)Lab anechoic chamber (range: 18 GHz to 110 GHz) is used to measure the 28 GHz and 39 GHz antennas.

Full size image
Fig. 16
figure 16

Comparison of simulated and measured (a) frequency vs. gain responses, (b) simulated 3D gain for \(G_1\) antenna at 750 MHz, and (c) measured 3D gain for \(G_1\) antenna at 750 MHz.

Full size image
Fig. 17
figure 17

Comparison of simulated and measured (a) frequency vs. gain responses, (b) simulated 3D gain for \(G_2\) antenna at 1700 MHz, and (c) measured 3D gain for \(G_2\) antenna at 1700 MHz.

Full size image
Fig. 18
figure 18

Comparison of simulated and measured (a) frequency vs. gain responses, (b) simulated 3D gain for \(G_3\) antenna at 1900 MHz, and (c) measured 3D gain for \(G_3\) antenna at 1900 MHz.

Full size image
Fig. 19
figure 19

Comparison of simulated and measured (a) frequency vs. gain responses, and (b) simulated 3D gain for \(G_4\) antenna at 3500 MHz, and (c) measured 3D gain for \(G_4\) antenna at 3500 MHz.

Full size image
Fig. 20
figure 20

Comparison of simulated and measured (a) frequency vs. gain responses, (b) simulated 3D gain for \(G_5\) antenna at 5200 MHz, and (c) measured 3D gain for \(G_5\) antenna at 5200 MHz.

Full size image
Fig. 21
figure 21

Comparison of frequency vs gain and radiation pattern, responses at 28 GHz and 39 GHz.

Full size image
Table 2 Performance comparison of our antenna with previously reported aperture shared antenna.
Full size table

Discussion

A compact, low-profile, co-integrated antenna is designed, fabricated, and characterized to operate in seven different frequency bands, namely 600 MHz to 900 MHz, 1500 MHz to 1800 MHz, 1800 MHz to 2700 MHz, 3500 MHz to 3700 MHz, 4700 MHz to 6000 MHz, as well as 28 GHz and 39 GHz. The low-frequency bands provide long-range coverage, while mmWave bands provide robust high-speed connectivity using beam steerability. The novel co-integrated antenna employs Inv-L and patch antennas for low-frequency operations and aperture-coupled feed patch antenna arrays for mmWave frequencies. The measured results match the simulated results very well for both the sub-6 GHz and mmWave bands. All the antennas on the aperture showed very good crosstalk performance as well. The isolation among the antennas can be further enhanced by using through-via fencing around the elements or by choosing different types of antennas. The mmWave antennas are tested only for a fixed beam application. The beam steerability can be added for 5G/6G or for later technology compatibility. The results demonstrate broad frequency coverage, stable realized gain, and consistently low crosstalk (below –20 dB) within a compact 200 mm \(\times\) 85 mm footprint, making the antenna suitable for applications such as fixed wireless access, small cells, integrated access and backhaul nodes, vehicular communication, satellite systems, and defense platforms. The beam steerability can be added for 5G/6G or for later technology compatibility. Looking ahead, future improvements include developing single antennas that cover multiple adjacent bands to simplify the aperture, integrating multiple multiband antennas into a single compact board to reduce size, enhancing isolation using techniques such as via fencing, decoupling structures, or metasurface integration, and extending the concept to support reconfigurability and beam-steerable mmWave arrays for next-generation communication systems.

Method

Full-wave simulation software ANSYS-HFSS was used to design the seven-band antenna. The design involved incorporating seven different bands of antenna into a single platform. A prototype was fabricated and tested to validate the full-wave simulation results. The seven-band antenna prototype is manufactured using a conventional laser machine. To implement the millimeter-wave antennas, they were fabricated in-house using the laser printing machine.